DICHROMATIC VISION 165 



The method of procedure was as follows, v. Helmholtz' spectro- 

 photometer was used, one half of the field being illuminated with the 

 homogeneous light, the other half with the mixture. In this apparatus 

 the lights are derived from two sources. The mixture is obtained 

 from one source by forming two spectra with the aid of a doubly 

 refracting prism, so that these spectra are polarised at right angles 

 to each other. The intensities of the lights in both halves of the field 

 are altered by varying the widths of the slits transmitting the lights 

 and by means of Nicol prisms. 



Suppose that a deuteranope is being examined and that the mixing 

 lights are 645 yn/x and 460-8 /x/x. One half of the field is illuminated 

 with a homogeneous red, say 670-8 /x/x. With this light a perfect 

 match, both in colour and brightness, can be made with 645 ju/x merely 

 by altering the intensity and without adding any of the 460-8 /x/x light. 

 The red or warm {W-) value of 670-8 fxfM is expressed in arbitrary units, 

 which depend upon the width of the slit, which transmits this light 

 from the second source of light, when the two exactly match. Let us 

 suppose that it equals 33 under the given conditions. 



A second wave-length is then chosen, e.g. 656 /XjU, the width of the 

 slit which transmits this light from the first source of light being kept 

 unchanged. We now find that the 645 /x/x slit must be opened more 

 in order to establish a perfect match, say to 48. For each succeeding 

 wave-length this slit must be opened wider until about 600 /x/x is 

 reached. Beyond this point the slit-widths must be reduced. 



At about 540 /x/x it is found that no perfect match can be made 

 with the 645 /x/x light alone. For example, 536 /x/x appears to the 

 deuteranope less saturated than 645 /t/x. It is necessary to add a 

 small quantity of light of shorter wave-length. On adding the appro- 

 priate amount of 460-8 /x/x light a perfect match, both in colour and 

 brightness, is obtained. This light, 536 /x/x, therefore possesses both 

 a red or IF-value and a blue or cold {K-) value, and these are recorded, 

 viz.. If -value 41, Z-value 6-3. 



The readings are not quite so simple as would appear from this 

 description. For example, it is found necessary to alter the width of 

 the slit which transmits the homogeneous light, otherwise the second 

 slit has to be opened so wide that the two spectra derived from it are 

 impure. Moreover the relative amounts of 645 /x/x and 460-8 /x/x lights 

 mixed, as well as the brightness of the resultant mixture, are altered 

 by rotation of the Nicol prism, and corrections have to be applied 

 according to the angle of rotation. 



